Skip to content

Commit

Permalink
Update docstrings and PEP-8 formatting in globaluncertainty.py
Browse files Browse the repository at this point in the history
  • Loading branch information
@mliu49
mliu49 committed Dec 9, 2019
1 parent c9bfa96 commit 924a1c4
Showing 1 changed file with 36 additions and 30 deletions.
66 changes: 36 additions & 30 deletions rmgpy/tools/globaluncertainty.py
View file
Original file line number Diff line number Diff line change
Expand Up @@ -33,8 +33,8 @@

import numpy as np

import muq.Modeling as muqm
import muq.Approximation as muqa
import muq.Modeling as muqm
import muq.Utilities as muqu


Expand All @@ -47,27 +47,31 @@ def __init__(self, cantera, output_species_list, k_params, k_uncertainty, g_para
======================= ====================================================
`cantera` A Cantera() object containing CanteraConditions and initialized species and reactions
`output_species_list` A list of Species() objects corresponding to the desired observables for uncertainty analysis
`k_params` Uncorrelated: A list of indices of the Reaction() objects in cantera.reaction_list corresponding to the uncertain input rate coefficients
Correlated: this is a list of strings of the uncertain kinetic parameter sources to be propagated in the model. i.e. 'H_Abstraction CHO/Oa'
`k_uncertainty` Uncorrelated: A list of uncertainties dlnk corresponding to the reactions in kReactions
Correlated: A list of dictionaries corresponding to each reaction's partial uncertainties with respect to various kinetic sources.
This is the output object from the Uncertainty.kineticInputUncertainties
`g_params ` Uncorrelated: A list of indices for the Species() objects in cantera.species_list corresponding to the uncertain input free energies of individual species
Correlated: A list of strings corresponding to the uncertain thermo sources to be propagated. i.e. 'Group(group) C=O'
`g_uncertainty` Uncorrelated: A list of uncertainties dG corresponding to the species in gSpecies in units of kcal/mol
Correlated: A list of dictionaries corresponding to each specie's partial uncertainties with respect to various thermo sources.
This is the output object from the Uncertainty.thermoInputUncertainties
`correlated` A flag set to either `True` or `False` depending on whether the uncertainties are correlated or not
If correlated, the correlated uncertainties will propagate and affect multiple species thermo and reaction kinetics parameters
`affected_reactions` Used only in the correlated case, this is a list of the indices of the affected rxns for k_params
`affected_species` Used only in the correlated case, this is a list of the indices of the affected species for g_params
`k_params` Uncorrelated: A list of RMG reaction indices corresponding to the uncertain
input rate coefficients
Correlated: A list of strings of the uncertain kinetic parameter sources to be
propagated in the model. i.e. 'H_Abstraction CHO/Oa'
`k_uncertainty` Uncorrelated: A list of uncertainties dlnk for every reaction
Correlated: A list of dictionaries with every reaction's partial uncertainties
with respect to various kinetic sources
This is the output object from the Uncertainty.kineticInputUncertainties
`g_params ` Uncorrelated: A list of RMG species indices corresponding to the uncertain input free energies
Correlated: A list of strings corresponding to the uncertain thermo sources to be
propagated in the model. i.e. 'Group(group) C=O'
`g_uncertainty` Uncorrelated: A list of uncertainties dG corresponding for every species in units of kcal/mol
Correlated: A list of dictionaries with every species' partial uncertainties
with respect to various thermo sources.
This is the output object from the Uncertainty.thermoInputUncertainties
`correlated` If true, propagate correlated uncertainties from the underlying independent parameters
`affected_reactions` For correlated only, a list of the indices of the affected rxns for k_params
`affected_species` For correlated only, a list of the indices of the affected species for g_params
`logx` A flag to set whether to use mole fraction or ln(mole fraction) as the output variable
============================================================================
"""
self.cantera = cantera
self.output_species_list = output_species_list
self.output_species_indices = [cantera.species_list.index(output_species) for output_species in output_species_list]
self.output_species_indices = [cantera.species_list.index(output_species)
for output_species in output_species_list]
self.correlated = correlated
self.logx = logx

Expand Down Expand Up @@ -98,24 +102,26 @@ def __init__(self, cantera, output_species_list, k_params, k_uncertainty, g_para
g_params = new_g_params

# for uncorrelated case, these are indices of reactions
# for correlated case, this is the list of labels for the uncertain parameters to be perturbed, i.e. 'H_Abstraction CHO/Oa'
# for correlated case, this is a list of parameter labels, i.e. 'H_Abstraction CHO/Oa'
self.k_params = k_params
# for uncorrelated case, these are indices of the species
# for correlated case, this is the list of labels for the uncertain thermo parameters to be perturbed, i.e. 'Group(ring) cyclohexane'
# for correlated case, this is a list of parameter labels, i.e. 'Group(ring) cyclohexane'
self.g_params = g_params

if not self.correlated:
k_uncertainty_factors = [val * np.sqrt(3) / np.log(10) for val in k_uncertainty]
self.k_uncertainty_factors = {}
for i, rxn_index in enumerate(k_params):
self.k_uncertainty_factors[rxn_index] = k_uncertainty_factors[rxn_index]
logging.debug('For {0}, set uncertainty factor to {1}'.format(cantera.reaction_list[rxn_index], k_uncertainty_factors[rxn_index]))
logging.debug('For {0}, set uncertainty factor to {1}'.format(cantera.reaction_list[rxn_index],
k_uncertainty_factors[rxn_index]))

g_uncertainty_factors = [val * np.sqrt(3) for val in g_uncertainty]
self.g_uncertainty_factors = {}
for i, spc_index in enumerate(g_params):
self.g_uncertainty_factors[spc_index] = g_uncertainty_factors[spc_index]
logging.debug('For {0}, set uncertainty factor to {1}'.format(cantera.species_list[spc_index], g_uncertainty_factors[spc_index]))
logging.debug('For {0}, set uncertainty factor to {1}'.format(cantera.species_list[spc_index],
g_uncertainty_factors[spc_index]))

else:
# In the correlated case, keep track of which reactions and species each
Expand All @@ -130,7 +136,7 @@ def __init__(self, cantera, output_species_list, k_params, k_uncertainty, g_para
uncertainty_factor = rxn_input_dict[k_param] * np.sqrt(3) / np.log(10)
# If this parameter string contributes to the reaction, add this reaction index
rxn_partial_uncertainty.append((rxn_index, uncertainty_factor))
self.k_uncertainty_factors[k_param] = rxn_partial_uncertainty # list of indices of the reactions affected
self.k_uncertainty_factors[k_param] = rxn_partial_uncertainty # list of indices of affected reactions
self.affected_reactions = list(affected_reactions)

self.g_uncertainty_factors = {}
Expand Down Expand Up @@ -177,7 +183,7 @@ def EvaluateImpl(self, inputs):
g_rv = inputs[0][len(self.k_params):]

if not self.correlated:
# Make deepcopies of the thermo and kinetics so as to not modify the originals in the species_list and reaction_list
# Make deepcopies of the thermo and kinetics to not modify originals in the species_list and reaction_list
original_thermo = [copy.deepcopy(self.cantera.species_list[index].thermo) for index in self.g_params]
original_kinetics = [copy.deepcopy(self.cantera.reaction_list[index].kinetics) for index in self.k_params]

Expand All @@ -190,7 +196,7 @@ def EvaluateImpl(self, inputs):
self.scale_to_thermo(rv, self.g_uncertainty_factors[spc_index], spc_index)

else:
# Make deepcopies of the thermo and kinetics so as to not modify the originals in the species_list and reaction_list
# Make deepcopies of the thermo and kinetics to not modify originals in the species_list and reaction_list
original_kinetics = [copy.deepcopy(self.cantera.reaction_list[index].kinetics)
for index in self.affected_reactions]
original_thermo = [copy.deepcopy(self.cantera.species_list[index].thermo)
Expand Down Expand Up @@ -302,7 +308,7 @@ class ReactorPCEFactory(object):
2. Run local uncertainty analysis
3. Extract top N most uncertain parameters
4. Input these set of parameters into a MUQ model class (which is a child of the ModPiece class)
5. Create EvaluateImpl function within model class that runs simulation based on reactor conditions through Cantera
5. Create EvaluateImpl function within model class that runs Cantera simulation based on reactor conditions
6. Perform PCE analysis of desired outputs
"""

Expand All @@ -320,11 +326,11 @@ def __init__(self, cantera, output_species_list, k_params, k_uncertainty, g_para
logx=logx,
)

# Select the polynomial and quadrature families
quad_family = muqa.GaussPattersonQuadrature() # We select the Gauss-Patterson quadrature as it is recommended
# as the fastest in the Patrick Conrad, Youssef Marzouk paper
poly_family = muqa.Legendre() # Uniform random variables used chemical kinetics
# uncertainty propagation uses Legendre polynomials
# We select the Gauss-Patterson quadrature as it is recommended as the fastest in the
# Patrick Conrad, Youssef Marzouk paper doi: 10.1137/120890715
quad_family = muqa.GaussPattersonQuadrature()
# Uniform random variables used chemical kinetics uncertainty propagation uses Legendre polynomials
poly_family = muqa.Legendre()

# Initialize the PCE Factory
self.dim = self.reactor_mod.inputSizes[0]
Expand Down

0 comments on commit 924a1c4

Please sign in to comment.